Total Synthesis of the Photorhabdus temperata ssp. Cinereal 3240 Zwitterionic Trisaccharide Repeating Unit

Abstract

Zwitterionic carbohydrate modifications, such as phosphoethanolamine (PEtN), govern host-pathogen interactions. While it is recognized that these modifications stimulate the host immune system, the purpose of PEtN modification remains largely descriptive. As an enabling step toward studying this carbohydrate modification, we report a total synthesis of the P. temperata zwitterionic trisaccharide repeating unit. The 32-step synthesis was enabled by H-phosphonate chemistry to install the PEtN arm on a poorly reactive and sterically hindered C4-alcohol.

Graphical Abstract

INTRODUCTION

Heterorhabditis is a genus of free-living nematode (Figure 1) of which all species are obligate parasites of insects.¹ In their earliest form, the nematode exists as an infective juvenile (IJ) that is colonized by Photorhabdus temperata (P. temperata), a gram-negative rod. P. temperata is mutualistically associated with entomopathogenic nematodes. After infection of an insect, the IJ releases P. temperata which multiples and destroys the insect. P. temperata also converts the insect host to a nutrient source for the reproducing nematode. Once the cadaver is depleted and dry, mature nematodes emerge to find new insect hosts. Interestingly, worm species differ in their ability to survive this type of arid environment. We hypothesize that one mechanism responsible for desiccation survival is production of exopolysaccharides by the symbiont. While exopolysaccharides protect the bacterial cell, we hypothesize they also increase water retention for the nematode.

Figure 1.

The nematode life cycle and its symbiont’s cell surface zwitterionic polysaccharide.

Recently, a novel zwitterionic polysaccharide has been identified on the cell surface of P. temperata (Figure 1).² Nematodes affiliated with this microorganism showcase enhanced survival during their emergence from a cadaver. To characterize why production of such disparate exopolysaccharides differentially affects the survival of host organisms, we initiated a campaign to synthesize the zwitterionic repeating unit. Structurally, the trisaccharide repeating unit 1 (Scheme 1) is composed of glucuronic acid (GlcA), N-Acetylgalactosamine (GalNAc), and the bacterial sugar 2-acetamido-4-amino-2,4,6-trideoxy-d-galactopyranose (AAT). From an analytical purview, the molecule incorporates all 1,2-trans-β-glycosidic bonds. Typically, anchimeric assistance from an acyl group at C2 governs diastereoselectivity during this type of glycosidic bond formation. However, the strategy is difficult when the glycosidic bond is to galactosamine based residues.⁴ Thus, proper masking of the C2 amine to control diastereoselectivity, and unmasking to reveal the native amide, must be correctly orchestrated. Lastly, we anticipated that installing phosphoethanolamine (PEtN, 2) on the poorly nucleophilic and sterically hindered axial alcohol of the GalNAc residue 3 would be challenging.

Scheme 1.

First-Generation Analysis and the Synthesis of an AAT donor.^a

^aReagents and conditions: (a) I₂ (1 equiv.), PPh₃ (1 equiv.), ImH (2.5 equiv.), THF, 23 °C, 4 h, 85%; (b) NaBH₃CN (5 equiv.), DMF, 100 °C, 30 h, 80%; (c) BzCl (1 equiv), pyr., −30 °C, 91% (d) Tf₂O (1.2 equiv.), pyr. (2 equiv.), CH₂Cl₂, 0 °C, 1 h then NaN₃ (3 equiv.), DMF, 23 °C, 3 h, 80%; (e) CAN (5 equiv.), CH₃CN/H₂O (3:1), 0 °C, 3 h then CCl₃CN (5 equiv.), DBU (0.3 equiv.), CH₂Cl₂, 0 °C, 1 h, 70%.

Based on this preliminary evaluation, the first-generation approach to the repeating unit centered on synthesis of frame 4 where the PEtN modified residue is on the non-reducing end to alleviate steric hindrance about the C4 alcohol (Scheme 1). As variations of building blocks 6 and 7 are known, the project started with the synthesis of 15 from triol 9.³ Functional group interconversion to the 6-iodo glucoside 10 was followed by a NaBH₃CN mediated reduction, giving 11 in 80% yield.^{3, 4} Next, regioselective benzoylation of O-3 gave alcohol 12,⁵ which was converted to its triflate, and displaced with NaN₃ to give 13.⁶ Lastly, oxidative cleavage of the PMP acetal with cerium ammonium nitrate (CAN) and conversion of the resulting alcohol to its trichloroacetimidate gave 15, primarily as the α-anomer, in 70% yield.⁷ While this synthesis of AAT is not radically different from established routes,^8–11 we highlight that the sequence is operable at gram scale and features minimal chromatographic separations.

With ATT secured, we next studied the coupling between known building blocks 16¹² and 17¹³ under NIS/TMSOTf mediated conditions (Scheme 2). In an orienting experiment we obtained disaccharide 18 in 65% yield. Although we anticipated high β-selectivity, due to anchimeric assistance from the C2 trichloroacetamide, we were unnerved to have isolated the α-anomer as the sole diastereomer. While we initially suspected that selectivity was guided by a torsional effect from the 4,6-O-benzylidene acetal,¹⁴ modification to donor 19 (i.e. exchange of the silyl ether for an acetate) generated the desired β-anomer 20 in 70% yield.

Scheme 2. C3 Directed Glycosylation ^a.

^aReagents and conditions (B): (a) NIS (2 equiv.), TMSOTf (0.15 equiv.), CH₂Cl₂, 4 Å MS, −78 °C, 1 h, α-only, 65%; (b) NIS (2 equiv.), TMSOTf (0.15 equiv.), CH₂Cl₂, 4 Å MS, −78 °C, 1.5 h, β-only, 70%.

While protecting groups prevent undesired sites from competing during a reaction, it is well-known that they can have a profound effect on the reactivity of the entire molecule. In a seminal example, Fraser-Reid’s team observed that 4,6-O-benzylidene acetals have a disarming (retarding oxocarbenium ion formation) effect on the hydrolysis of O-pentenyl glycosides.^{15, 16} This effect was attributed to increased torsional strain in the fused bicyclic system as the chair−chair glycosyl donor collapses to an intermediate chair−boat oxocarbenium ion. Unfortunately, applying this logic to the selectivity observed in the production of 18 and 20 is flawed as both donors 16 and 19 incorporate a benzylidene acetal protecting group. Thus, we figured that an electronic element from C3 was governing diastereoselectivity.

Our mechanistic hypothesis starts with the α-selective pathway (Scheme 2).^17–20 After NIS/TMSOTf mediated activation of thioglycoside 16, we hypothesize that an electron donating silyl ether at C3 would destabilize the intermediate glycosyl triflate 21. As a consequence, we anticipate a shift in the covalent α-glycosyl triflate 21 − solvent-separated ion pair (SSIP) 22 equilibria to favor an α-selective SSIP. In the β-selective case, the result is best interpreted as the C3 acetate imparting a moderate electron-withdrawing effect which acts to increase the energetic barrier between a covalent glycosyl triflate and the oxocarbenium ion, resulting in a diminished concentration of the SSIP. Consequently, one would anticipate the reactive intermediate to exist in the form of α-glycosyl triflate 24, a contact ion pair (CIP) 25, or an oxocarbenium ion 26 stabilized through anchimeric assistance. In each case, increased β-selectivity would be expected. We add that this analysis is hypothetical and spectroscopic studies are ongoing to characterize the pathways involved in producing the observed selectivity’s.

Returning to the synthesis (Scheme 3), 20 was exposed to CAN to liberate the anomeric alcohol. Following conversion to the trichloroacetimidate 27, we were surprised by our inability to unite donor 27 with AAT acceptor 14 under a multitude or Lewis and Bronsted acidic reaction conditions. It is well known that uronic acid glycosyl donors perform poorly in glycosylation reactions due to the C6 ester which destabilizes oxocarbenium ions. We were hopeful, however, that use of trichloroacetimidate donors, as demonstrated elegantly by Boons and co-workers would enable productive glycosylation.²¹ While we quickly learned that the use of ester protecting groups at C2 and C3 would also render the Boons approach intractable, we were surprised that exchange of the benzoate protecting groups for electron rich benzyl ethers did not solve the problem. Accordingly, we developed a 2^nd generation approach (Scheme 4).

Scheme 3. Failed synthesis of the core trisaccharide. ^a.

^aReagents and conditions: (a) CAN (5 equiv.), CH₃CN/H₂O (3:1), 0 °C, 3 h then CCl₃CN (5 equiv.), DBU (0.3 equiv.), CH₂Cl₂, 0 °C, 1 h, α-only, 60%.

Scheme 4. Successful second-generation approach and completion of the total synthesis ^a.

^aReagents and conditions: (a) NIS (2 equiv.), TfOH (0.15 equiv.), CH₃CN/CH₂Cl₂, 4 Å MS, −78 °C, 2 h, β/α (4:1), 91%; (b) 1.5M NaOCH₃, CH₃OH, 23 °C, 1 h, >95%; (c) TfOH (0.15 equiv.), CH₂Cl₂, 4 Å MS, −78 °C, 1 h, β-only, 90%; (d) Et₃SiH (3 equiv.), TfOH (3.5 equiv.), CH₂Cl₂, 4 Å MS, −78 °C, 2 h, 83%; (e) PivCl (2.0 equiv.), pyr., 23 °C, 12 h then iodine (2.5 equiv.), pyr/H₂O (19:1), 23 °C, 6 h, 81%; (f) Pd(OH)₂/C (7 equiv.), H₂ (balloon), CH₃OH, AcOH, 66 h then THF, 1.5M NaOCH₃, CH₃OH/H₂O, 23 °C, 12 h, 65%.

A defining phase of the 2^nd generation approach involved recourse to a frame shift wherein the AAT building block was moved to the non-reducing end of the repeating unit 29, a maneuver which jettisoned the glucuronic acid residue as both a donor and acceptor. While we had addressed the glucuronic acid glycosylation issue, we would now need to install the phosphorus arm on a very sterically hindered alcohol.

Undeterred, we moved forward with this 2^nd generation route by coupling thioglycoside 30 with glucuronate acceptor 31 at −78 °C using TMSOTf as the promoter. The reaction provided disaccharide 32 with complete β-selectivity, albeit in a deplorable 48% yield. The remainder of the gram mass was unreacted starting material. Evaluation of the system showed that extended reaction times would not significantly improve the yield. While conducting the coupling at warmer temperatures was significantly more productive (72%), diastereoselectivity suffered, dropping to 2:1. Ultimately, we were able to access a tractable route to 32 by using TfOH as the promoter at −78 °C in a CH₂Cl₂:CH₃CN solvent system with a 4:1 - β:α ratio in 91% yield. Moving forward, disaccharide 32 was subjected to Zemplén conditions to obtain 33 in near quantitative yield. With acceptor 33 in hand, glycosylation with the trichloroacetimidate donor 15 at −78 °C using TfOH as the promoter gave 34 exclusively as the β-anomer in 90% yield. At this stage, the synthesis required reductive opening of the benzylidene acetal – a reaction that, while common, is far from trivial.²² After careful optimization, an Et₃SiH - TfOH reducing system provided the reductive benzylidene acetal ring opening with the desired regioselectivity.^{23, 24} Accordingly, trisaccharide 35 was isolated with a free axial C4’’ alcohol in 83% yield.

Having prepared the desired key intermediate, there were two a priori considerations to install the phosphorus arm. Given that the acceptor is located on the central residue, is in a 1,2-cis relationship with the terminal monosaccharide, and is axial we anticipated phosphorylation would be difficult. Accordingly, our approach to phosphorylation focused on reaction with phosphoramidite 38, which resides in the highly electrophilic phosphorus (III) oxidation state. Indeed, previous experience from our lab has shown that electrophiles of type 38 are useful reagents to install phosphocholine onto hindered nucleophiles.^{12, 25} Unfortunately, the coupling of phosphoramidite 38 with 36 (under a range of reaction conditions), using 1H-tetrazole as the activator failed to form the O-P bond in reasonable yield. Indeed, the only productive reaction (33% yield) observed in our early studies was when 4,5-dichloroimidazole (DCI) was employed as the activator to form compounds of type 40.

In need of a more efficient coupling process, we turned to the synthetic oligonucleotide literature and discovered that a common tactic to generate O-P bonds on sterically hindered or poorly nucleophilic residues is to leverage the reactivity of H-phosphonates.²⁶ We hypothesized that we could use reagent 39 to install the phosphoethanolamine residue. H-phosphonates work according to the following mechanism. First, the phosphonate is activated by pivaloyl chloride, which generates a system in which the electron withdrawing character of the pivaloyl group makes the phosphorus(V) center sufficiently electrophilic. Accordingly, upon reaction with an alcohol nucleophile, rapid transesterification leads to formation of the O-P bond. At this stage oxidation then leads to the desired motif. Gratifyingly, coupling between 35 and 39 was indeed executed in the presence of pivaloyl chloride and pyridine to form the O-P bond. Next, oxidation with I₂ provided the phosphorylated trisaccharide 36 in 81% yield.

With the fully masked repeating unit in hand, we commenced with global deprotection. First, exhaustive hydrogenolysis over Pearlman’s catalyst was used to remove a benzyl carbamate and three benzyl ethers. The reducing environment also reduced one azide to its amine, and two trichloroacetamide groups to their corresponding acetamide. After a short column to remove organic soluble impurities, the material was carried forward and exposed to NaOCH₃ in aq. CH₃OH at ambient temperature to remove a single benzoate protecting group and saponify the methyl ester. Following purification of the crude-material using size exclusion chromatography, 29 was isolated as a white solid. Characterization of the repeating unit using ¹H, ¹³C, ³¹P NMR provided comparable spectra to the native polymer (see Table S1 in the supplemental information).

CONCLUSION

Reported herein is the first total synthesis of the Photorhabdus temperata ssp. Cinereal 3240 Zwitterionic Trisaccharide Repeating Unit. The synthesis highlights a scalable synthesis of AAT and the use of an H-phosphonate donor to install phosphoethanolamine functionality on a sterically hindered, poorly-nucleophilic alcohol. Our pursuit toward the synthesis of additional ZPS of interest is ongoing, as is the conversion of these synthetic materials to tool compounds to characterize why microorganisms use zwitterionic motifs in host-pathogen interactions.